XAI-KG@ESWC2025

ESWC 2025 - 1st International Workshop on

eXplainable AI and Knowledge Graphs

The synergy between eXplainable AI (XAI) and Knowledge Graphs (KGs) has gained momentum as an essential approach for achieving transparency, trust, and understanding in AI systems. Knowledge Graphs provide a structured, interconnected framework for representing domain-specific knowledge, while XAI aims either to provide insight for predicted results or to clarify how machine learning models function internally, particularly deep learning systems, which are often complex and difficult to interpret.  By leveraging KGs within XAI, researchers and practitioners can enhance the understanding and interpretability of AI models, enabling explanations that are both contextual and relevant to domain knowledge, making it easier for users to trust and understand AI-driven insights and decisions.

The combination of XAI and KGs presents unique advantages and challenges. KGs can serve as an intuitive map for AI reasoning paths, offering insights into the relationships and logic that AI systems use to reach conclusions. This can be particularly valuable in applications requiring high levels of transparency, such as healthcare, finance, and law, where understanding the rationale behind AI predictions and actions is crucial. Conversely, XAI can assist in constructing and refining KGs, helping to identify which aspects of a graph's structure contribute most to accurate, reliable reasoning, ultimately enriching KG content with a layer of explainable intelligence.

This workshop aims to bring together researchers, practitioners, and industry experts to explore the vast opportunities and specific challenges of combining XAI with KGs. We invite discussion on novel methodologies, applications, and case studies demonstrating how KGs can improve interpretability in complex AI models, and how XAI can, in turn, enhance knowledge extraction, inference, and reasoning within KGs. Topics will span theoretical advances, practical tools, and industry applications, fostering dialogue on how KGs can make black-box AI systems more understandable, and how explainability can guide KG development.